Uriel
03-03-2016, 12:25 AM
It's a relatively old article (January 20, 2016), but most of its arguments still hold true today.
Ranking the titans in basketball
Spurring on the Warriors
Jan 20th 2016, 15:16 BY D.R.
http://cdn.static-economist.com/sites/default/files/imagecache/full-width/images/2016/01/blogs/game-theory/20160123_blp516.jpg
IT IS a truth universally acknowledged that the story of the 2015-16 National Basketball Association (NBA) season is the story of the Golden State Warriors. The public should have already focused its attention on the club last year, when they became the tenth team ever to win at least 67 of 82 regular-season contests, and marched through the playoffs to their first championship in 40 years without ever facing an elimination game. But the greatest source of intrigue for fans in 2014-15 was the return (http://www.economist.com/blogs/gametheory/2014/07/basketball-economics) of LeBron James (http://www.lebronjames.com/), the sport’s best player, to his hometown Cleveland Cavaliers after spending four years and winning two titles with the Miami Heat. And the most interesting aspect of the Finals was Mr James’s valiant but vain attempt to carry his club on his back singlehandedly (http://www.economist.com/blogs/gametheory/2015/06/individual-v-team-basketball) after both of his co-stars went down with injuries.
Only during this season, which just passed its halfway mark, are the Warriors getting their due. The team won its first 24 games in a row, setting a new record to start a campaign, and are currently 39-4, putting them on pace to best Michael Jordan (http://en.wikipedia.org/wiki/Michael_Jordan)’s 1995-96 Chicago Bulls for the best regular-season performance in NBA history. Moreover, they are winning in the most visually compelling style since perhaps the “Showtime” Los Angeles Lakers of the 1980s, an up-tempo carnival of long-range sharpshooting (https://www.youtube.com/watch?v=X9necA_prVM), transition baskets (https://www.youtube.com/watch?v=I9UnipgjQMs) and suffocating defence (https://www.youtube.com/watch?v=3MgeMoE0EXo). Stephen Curry (https://en.wikipedia.org/wiki/Stephen_Curry), their electric superstar point guard, has supplanted Mr James as the face of the league, inspiring odes to his ballet-like artistry (http://www.nytimes.com/2015/11/25/sports/basketball/artistry-of-stephen-curry-with-golden-state-warriors.html) as well as crunchy statistical analyses (http://fivethirtyeight.com/features/stephen-currys-bombs-are-too-good-to-be-true) crowning him a one-man revolution (http://fivethirtyeight.com/features/stephen-curry-is-the-revolution). And the Warriors are singlehandedly propping up the NBA’s finances, vastly boosting attendance (http://fivethirtyeight.com/features/the-warriors-are-the-hottest-ticket-in-every-town) in every city they visit. The indispensable Basketball-Reference.com (http://www.basketball-reference.com/) has even placed a prominent highlight box dedicated to “2015-16 Warriors Feats and Records” on its homepage.
There’s only one catch to the narrative about the Year of the Warriors: they’re not the best team in the league. No, really. The case against Golden State is surprisingly simple. The goal of a basketball team is to win, and the way you win is by outscoring the other guy. The Warriors have indeed done just that, averaging 12.1 more points per game than their opponents. That is an outstanding mark: in the past only Mr Jordan’s best Bulls team, the 1970-71 Milwaukee Bucks (featuring the all-time greats Kareem Abdul-Jabbar (http://www.kareemabduljabbar.com/) and Oscar Robertson (http://www.thebigo.com/)) and the 1971-72 Los Angeles Lakers with their four Hall of Famers (http://www.hoophall.com/) have exceeded it. However, another team in the current season is on pace to smash the Lakers’ record of 12.3: the San Antonio Spurs, who happened to win a title (their fifth over the past 17 seasons) the year before the Warriors did. Although they have won a lower percentage of their games than the Warriors have (85.7% to 90.7%), this year’s edition of the club is outscoring its opponents by a whopping 14.2 points per contest. At 2.1 points per game, the gap between the Spurs’ and Warriors’ scoring margins is as big as the difference between a league-average team and one that would be expected to win around 57% of the time—which is a record consistent with a middling playoff seed and an outside shot (https://en.wikipedia.org/wiki/1994%E2%80%9395_Houston_Rockets_season) at a championship (https://en.wikipedia.org/wiki/1977%E2%80%9378_Washington_Bullets_season).
A sceptic might retort that there’s no trophy granted to the team with the best point differential. A club’s average margin of victory can certainly be misleading: some coaches choose to “run up the score” and rack up blowout wins even after a game is out of reach, while others prefer to rest their starters. In addition, if there is any truth to the notion that some players consistently “rise to the occasion” in the final minutes of close contests while others “choke”, then teams stacked with “clutch performers” should win more games than their scoring margins would indicate. However, the counterargument is that the distribution of a team’s total points scored and allowed within a season is largely random. As a result, a club’s overall output is likely to be a better indicator of future performance than its won-lost record.
To determine which of these theories is correct, I downloaded a dataset from Basketball-Reference containing the results of every regular-season NBA game between the advent of the three-point line in 1979 and the end of last season. Using the logistic regression (https://en.wikipedia.org/wiki/Logistic_regression)method, I derived a formula that predicts a team’s odds of winning a game, based on home-court advantage, the number of days since it last played, its year-to-date won-lost record and its year-to-date scoring margin, as well as the same figures for its opponent, its prior opponents earlier in the season, and its opponent’s prior opponents. Even though the model does not contain any information about the specific players on a team, injuries, or trades, and ignores everything that happened before the 2015-16 campaign (on opening day every team is given the same odds save for home-court advantage), it has performed reasonably well: for example, there have been 60 games so far this year when it assigned a team a probability of victory between 15% and 20%, and those clubs have prevailed 18% of the time.
http://cdn.static-economist.com/sites/default/files/imagecache/original-size/images/2016/01/blogs/game-theory/20160123_woc200.png
At first glance, the model suggests that both won-lost record and scoring margin contain distinctive, valuable information about team quality: rather than ignoring one or the other, it finds that both variables make a statistically significant contribution to its prediction. However, it is not shy about expressing a preference. After controlling for home court, rest days, and the quality of previous opponents, the difference between two teams’ scoring margins has nearly six times as large an impact on who is likely to win as the difference between their won-lost records (see chart).
Most of the time, the two measures move in tandem—after all, scoring more points is a great way to get more wins. However, they occasionally diverge, primarily because a team tends to win small and lose big, or the reverse. And the record is clear that when the two indicators point in opposite directions—as with the 1991-92 Indiana Pacers, who on January 30th were a mere 16-28 despite having outscored their opponents, or the 2003-04 Toronto Raptors, who started the season 11-8 while allowing 3.9 more points per game than they scored—it is point differential that usually points the right way. Those Pacers won 63% of their remaining games, while the Raptors prevailed in just 35%. The same applies to individual matchups: of the 18 contests since 1979-80 in which the won-lost and scoring-margin methods had the largest disagreement, the one preferred by scoring margin won 13 (see table).
http://cdn.static-economist.com/sites/default/files/imagecache/original-size/images/2016/01/blogs/game-theory/20160123_woc201.png
It’s hard to analyse the tactical specifics of a head-to-head matchup between the Warriors and Spurs, since they have not yet played each other: their first game will be must-see TV on January 25th. However, given the primacy of point differential, the surface evidence available suggests that San Antonio would be likely to beat Golden State in the playoffs. Without giving any advantages for home court or rest, the model sees the Spurs as a 58% favourite in a single game over the Warriors, which corresponds to a 67% chance of winning a seven-game series. That would narrow to around 62% (http://basketballnumbers.com/2010/09/06/a-sunday-kind-of-piece-special-labor-day-edition-parte-dos-building-probability-models-for-the-nba-playoffs) if Messrs. Curry and Co. secure home-court advantage. However, if the Spurs are really a better team, they should have a decent chance to make up their current 2.5-game deficit in the standings and become the top seed themselves. In that case, their chances of winning a seven-game series against Golden State would increase to 70%.
Is there any reason to doubt the model’s enthusiasm for the Spurs? Are they more likely than the Warriors to be playing over their heads, and thus likely to fall back to earth? They were certainly a far inferior team last year, when they won “only” 67% of their games and lost in the first round of the playoffs. That prompted many observers to suggest that their veteran trio of Tim Duncan (https://en.wikipedia.org/wiki/Tim_Duncan), Tony Parker (https://en.wikipedia.org/wiki/Tony_Parker) and Manu Ginobili (http://usa.manuginobili.com/) was on its last legs. But in response to that lacklustre showing, San Antonio made its flashiest acquisition in years by signing the free agent LaMarcus Aldridge (https://en.wikipedia.org/wiki/LaMarcus_Aldridge) (pictured), an accomplished 6’11” (2.11-metre) scorer and rebounder. They also picked up David West (https://en.wikipedia.org/wiki/David_West_(basketball)), a lockdown defensive forward who fits naturally with their team-first philosophy.
Now the Spurs essentially have two championship cores in one. By playing a slow, grinding pace (San Antonio ranks 23rd of the NBA’s 30 teams in possessions per game) and tightly limiting their minutes, Gregg Popovich (https://en.wikipedia.org/wiki/Gregg_Popovich), the Spurs’ coach, has found a fountain of youth for Messrs Duncan (39 years young), Ginobili (38) and Parker (33). All continue to excel when on the court—particularly Mr Duncan, an all-time inner-circle great, who is proving the adage about big men ageing gracefully by remaining the best defender in basketball (http://espn.go.com/nba/statistics/rpm/_/sort/DRPM). But they have all gracefully passed the baton of leadership to to 24-year-old Kawhi Leonard (https://en.wikipedia.org/wiki/Kawhi_Leonard)—a modestly regarded player in college, traded to the Spurs for the forgettable George Hill (https://en.wikipedia.org/wiki/George_Hill_(basketball)) immediately after being drafted, who has blossomed into a two-way super-duper-star (http://www.poundingtherock.com/2016/1/21/10806686/unique-kawhi-leonard-offense). And although Mr Aldridge is yet to set the world on fire as a Spur, he is precisely the kind of athletic big man San Antonio will need to keep pace with the Warriors’ “death squad (http://www.goldenstateofmind.com/2015/11/23/9777336/small-ball-death-squad)” lineup of five players 6’8” or shorter. Finally, although the Spurs have used him only sparingly—there is little room for experimentation on a team this successful—the 7’3” Serbian import Boban Marjanovic (https://en.wikipedia.org/wiki/Boban_Marjanovi%C4%87) has excelled in his first 180 minutes of play in the NBA, and could easily turn into a secret weapon come playoff time.
There is still a stronger case to keep expectations in check about San Antonio than there is about Golden State. The Warriors already proved last year that they can dominate the league wire-to-wire and vanquish Mr James; although the Spurs’ long-run record of success is second to none, they are yet to demonstrate they can win a title with Mr Leonard as their best player. No matter how well-rested Mr Popovich keeps his veterans, they could always tire or get injured as the season drags on. And there were plenty of doubts expressed about how well a pass-allergic player like Mr Aldridge would fit into the Spurs’ whirlwind of ball movement—or whether he was ever anywhere near as good as his reputation (http://www.boxscoregeeks.com/articles/lamarcus-aldridge-is-overrated) to begin with. The CARMELO (http://projects.fivethirtyeight.com/2016-nba-picks) forecasting system at FiveThirtyEight, which unlike the simple model used above does incorporate projections for individual players and their past performances, still sees the Warriors as a stronger team, with about a 54%-46% edge on a neutral court.
Nonetheless, such caveats can’t remotely account for the difference in attention paid to the two clubs so far this year. Perhaps it’s because all eyes were already on the Warriors following their 2014-15 title, or because of their 24-0 start, or because Mr Curry is the most exciting player to watch on television since Mr Jordan. Or maybe it’s because the Spurs have been great for 20 years so their success is no longer news, or because their commitment to depth and fundamentals leaves them without a single telegenic star, or because they started “slow” (they were “merely” 18-5 and 25-6) before their current 11-game win streak brought their won-lost record nearly into line with their scoring margin.
Regardless, there are two teams flirting with history this season, not one. And barring an unlikely upset earlier in the postseason, they are on a collision course for what promises to be an epic Western Conference finals. Las Vegas is currently giving the Warriors a better-than-even chance of winning the West (a bet of $100 would return an $80 profit if successful), while rewarding bettors who place their faith in the Spurs with a generous payout (the same $100 bet would yield a $150 gain). You can probably guess which side offers better value.
http://www.economist.com/blogs/gametheory/2016/01/ranking-titans-basketball
Ranking the titans in basketball
Spurring on the Warriors
Jan 20th 2016, 15:16 BY D.R.
http://cdn.static-economist.com/sites/default/files/imagecache/full-width/images/2016/01/blogs/game-theory/20160123_blp516.jpg
IT IS a truth universally acknowledged that the story of the 2015-16 National Basketball Association (NBA) season is the story of the Golden State Warriors. The public should have already focused its attention on the club last year, when they became the tenth team ever to win at least 67 of 82 regular-season contests, and marched through the playoffs to their first championship in 40 years without ever facing an elimination game. But the greatest source of intrigue for fans in 2014-15 was the return (http://www.economist.com/blogs/gametheory/2014/07/basketball-economics) of LeBron James (http://www.lebronjames.com/), the sport’s best player, to his hometown Cleveland Cavaliers after spending four years and winning two titles with the Miami Heat. And the most interesting aspect of the Finals was Mr James’s valiant but vain attempt to carry his club on his back singlehandedly (http://www.economist.com/blogs/gametheory/2015/06/individual-v-team-basketball) after both of his co-stars went down with injuries.
Only during this season, which just passed its halfway mark, are the Warriors getting their due. The team won its first 24 games in a row, setting a new record to start a campaign, and are currently 39-4, putting them on pace to best Michael Jordan (http://en.wikipedia.org/wiki/Michael_Jordan)’s 1995-96 Chicago Bulls for the best regular-season performance in NBA history. Moreover, they are winning in the most visually compelling style since perhaps the “Showtime” Los Angeles Lakers of the 1980s, an up-tempo carnival of long-range sharpshooting (https://www.youtube.com/watch?v=X9necA_prVM), transition baskets (https://www.youtube.com/watch?v=I9UnipgjQMs) and suffocating defence (https://www.youtube.com/watch?v=3MgeMoE0EXo). Stephen Curry (https://en.wikipedia.org/wiki/Stephen_Curry), their electric superstar point guard, has supplanted Mr James as the face of the league, inspiring odes to his ballet-like artistry (http://www.nytimes.com/2015/11/25/sports/basketball/artistry-of-stephen-curry-with-golden-state-warriors.html) as well as crunchy statistical analyses (http://fivethirtyeight.com/features/stephen-currys-bombs-are-too-good-to-be-true) crowning him a one-man revolution (http://fivethirtyeight.com/features/stephen-curry-is-the-revolution). And the Warriors are singlehandedly propping up the NBA’s finances, vastly boosting attendance (http://fivethirtyeight.com/features/the-warriors-are-the-hottest-ticket-in-every-town) in every city they visit. The indispensable Basketball-Reference.com (http://www.basketball-reference.com/) has even placed a prominent highlight box dedicated to “2015-16 Warriors Feats and Records” on its homepage.
There’s only one catch to the narrative about the Year of the Warriors: they’re not the best team in the league. No, really. The case against Golden State is surprisingly simple. The goal of a basketball team is to win, and the way you win is by outscoring the other guy. The Warriors have indeed done just that, averaging 12.1 more points per game than their opponents. That is an outstanding mark: in the past only Mr Jordan’s best Bulls team, the 1970-71 Milwaukee Bucks (featuring the all-time greats Kareem Abdul-Jabbar (http://www.kareemabduljabbar.com/) and Oscar Robertson (http://www.thebigo.com/)) and the 1971-72 Los Angeles Lakers with their four Hall of Famers (http://www.hoophall.com/) have exceeded it. However, another team in the current season is on pace to smash the Lakers’ record of 12.3: the San Antonio Spurs, who happened to win a title (their fifth over the past 17 seasons) the year before the Warriors did. Although they have won a lower percentage of their games than the Warriors have (85.7% to 90.7%), this year’s edition of the club is outscoring its opponents by a whopping 14.2 points per contest. At 2.1 points per game, the gap between the Spurs’ and Warriors’ scoring margins is as big as the difference between a league-average team and one that would be expected to win around 57% of the time—which is a record consistent with a middling playoff seed and an outside shot (https://en.wikipedia.org/wiki/1994%E2%80%9395_Houston_Rockets_season) at a championship (https://en.wikipedia.org/wiki/1977%E2%80%9378_Washington_Bullets_season).
A sceptic might retort that there’s no trophy granted to the team with the best point differential. A club’s average margin of victory can certainly be misleading: some coaches choose to “run up the score” and rack up blowout wins even after a game is out of reach, while others prefer to rest their starters. In addition, if there is any truth to the notion that some players consistently “rise to the occasion” in the final minutes of close contests while others “choke”, then teams stacked with “clutch performers” should win more games than their scoring margins would indicate. However, the counterargument is that the distribution of a team’s total points scored and allowed within a season is largely random. As a result, a club’s overall output is likely to be a better indicator of future performance than its won-lost record.
To determine which of these theories is correct, I downloaded a dataset from Basketball-Reference containing the results of every regular-season NBA game between the advent of the three-point line in 1979 and the end of last season. Using the logistic regression (https://en.wikipedia.org/wiki/Logistic_regression)method, I derived a formula that predicts a team’s odds of winning a game, based on home-court advantage, the number of days since it last played, its year-to-date won-lost record and its year-to-date scoring margin, as well as the same figures for its opponent, its prior opponents earlier in the season, and its opponent’s prior opponents. Even though the model does not contain any information about the specific players on a team, injuries, or trades, and ignores everything that happened before the 2015-16 campaign (on opening day every team is given the same odds save for home-court advantage), it has performed reasonably well: for example, there have been 60 games so far this year when it assigned a team a probability of victory between 15% and 20%, and those clubs have prevailed 18% of the time.
http://cdn.static-economist.com/sites/default/files/imagecache/original-size/images/2016/01/blogs/game-theory/20160123_woc200.png
At first glance, the model suggests that both won-lost record and scoring margin contain distinctive, valuable information about team quality: rather than ignoring one or the other, it finds that both variables make a statistically significant contribution to its prediction. However, it is not shy about expressing a preference. After controlling for home court, rest days, and the quality of previous opponents, the difference between two teams’ scoring margins has nearly six times as large an impact on who is likely to win as the difference between their won-lost records (see chart).
Most of the time, the two measures move in tandem—after all, scoring more points is a great way to get more wins. However, they occasionally diverge, primarily because a team tends to win small and lose big, or the reverse. And the record is clear that when the two indicators point in opposite directions—as with the 1991-92 Indiana Pacers, who on January 30th were a mere 16-28 despite having outscored their opponents, or the 2003-04 Toronto Raptors, who started the season 11-8 while allowing 3.9 more points per game than they scored—it is point differential that usually points the right way. Those Pacers won 63% of their remaining games, while the Raptors prevailed in just 35%. The same applies to individual matchups: of the 18 contests since 1979-80 in which the won-lost and scoring-margin methods had the largest disagreement, the one preferred by scoring margin won 13 (see table).
http://cdn.static-economist.com/sites/default/files/imagecache/original-size/images/2016/01/blogs/game-theory/20160123_woc201.png
It’s hard to analyse the tactical specifics of a head-to-head matchup between the Warriors and Spurs, since they have not yet played each other: their first game will be must-see TV on January 25th. However, given the primacy of point differential, the surface evidence available suggests that San Antonio would be likely to beat Golden State in the playoffs. Without giving any advantages for home court or rest, the model sees the Spurs as a 58% favourite in a single game over the Warriors, which corresponds to a 67% chance of winning a seven-game series. That would narrow to around 62% (http://basketballnumbers.com/2010/09/06/a-sunday-kind-of-piece-special-labor-day-edition-parte-dos-building-probability-models-for-the-nba-playoffs) if Messrs. Curry and Co. secure home-court advantage. However, if the Spurs are really a better team, they should have a decent chance to make up their current 2.5-game deficit in the standings and become the top seed themselves. In that case, their chances of winning a seven-game series against Golden State would increase to 70%.
Is there any reason to doubt the model’s enthusiasm for the Spurs? Are they more likely than the Warriors to be playing over their heads, and thus likely to fall back to earth? They were certainly a far inferior team last year, when they won “only” 67% of their games and lost in the first round of the playoffs. That prompted many observers to suggest that their veteran trio of Tim Duncan (https://en.wikipedia.org/wiki/Tim_Duncan), Tony Parker (https://en.wikipedia.org/wiki/Tony_Parker) and Manu Ginobili (http://usa.manuginobili.com/) was on its last legs. But in response to that lacklustre showing, San Antonio made its flashiest acquisition in years by signing the free agent LaMarcus Aldridge (https://en.wikipedia.org/wiki/LaMarcus_Aldridge) (pictured), an accomplished 6’11” (2.11-metre) scorer and rebounder. They also picked up David West (https://en.wikipedia.org/wiki/David_West_(basketball)), a lockdown defensive forward who fits naturally with their team-first philosophy.
Now the Spurs essentially have two championship cores in one. By playing a slow, grinding pace (San Antonio ranks 23rd of the NBA’s 30 teams in possessions per game) and tightly limiting their minutes, Gregg Popovich (https://en.wikipedia.org/wiki/Gregg_Popovich), the Spurs’ coach, has found a fountain of youth for Messrs Duncan (39 years young), Ginobili (38) and Parker (33). All continue to excel when on the court—particularly Mr Duncan, an all-time inner-circle great, who is proving the adage about big men ageing gracefully by remaining the best defender in basketball (http://espn.go.com/nba/statistics/rpm/_/sort/DRPM). But they have all gracefully passed the baton of leadership to to 24-year-old Kawhi Leonard (https://en.wikipedia.org/wiki/Kawhi_Leonard)—a modestly regarded player in college, traded to the Spurs for the forgettable George Hill (https://en.wikipedia.org/wiki/George_Hill_(basketball)) immediately after being drafted, who has blossomed into a two-way super-duper-star (http://www.poundingtherock.com/2016/1/21/10806686/unique-kawhi-leonard-offense). And although Mr Aldridge is yet to set the world on fire as a Spur, he is precisely the kind of athletic big man San Antonio will need to keep pace with the Warriors’ “death squad (http://www.goldenstateofmind.com/2015/11/23/9777336/small-ball-death-squad)” lineup of five players 6’8” or shorter. Finally, although the Spurs have used him only sparingly—there is little room for experimentation on a team this successful—the 7’3” Serbian import Boban Marjanovic (https://en.wikipedia.org/wiki/Boban_Marjanovi%C4%87) has excelled in his first 180 minutes of play in the NBA, and could easily turn into a secret weapon come playoff time.
There is still a stronger case to keep expectations in check about San Antonio than there is about Golden State. The Warriors already proved last year that they can dominate the league wire-to-wire and vanquish Mr James; although the Spurs’ long-run record of success is second to none, they are yet to demonstrate they can win a title with Mr Leonard as their best player. No matter how well-rested Mr Popovich keeps his veterans, they could always tire or get injured as the season drags on. And there were plenty of doubts expressed about how well a pass-allergic player like Mr Aldridge would fit into the Spurs’ whirlwind of ball movement—or whether he was ever anywhere near as good as his reputation (http://www.boxscoregeeks.com/articles/lamarcus-aldridge-is-overrated) to begin with. The CARMELO (http://projects.fivethirtyeight.com/2016-nba-picks) forecasting system at FiveThirtyEight, which unlike the simple model used above does incorporate projections for individual players and their past performances, still sees the Warriors as a stronger team, with about a 54%-46% edge on a neutral court.
Nonetheless, such caveats can’t remotely account for the difference in attention paid to the two clubs so far this year. Perhaps it’s because all eyes were already on the Warriors following their 2014-15 title, or because of their 24-0 start, or because Mr Curry is the most exciting player to watch on television since Mr Jordan. Or maybe it’s because the Spurs have been great for 20 years so their success is no longer news, or because their commitment to depth and fundamentals leaves them without a single telegenic star, or because they started “slow” (they were “merely” 18-5 and 25-6) before their current 11-game win streak brought their won-lost record nearly into line with their scoring margin.
Regardless, there are two teams flirting with history this season, not one. And barring an unlikely upset earlier in the postseason, they are on a collision course for what promises to be an epic Western Conference finals. Las Vegas is currently giving the Warriors a better-than-even chance of winning the West (a bet of $100 would return an $80 profit if successful), while rewarding bettors who place their faith in the Spurs with a generous payout (the same $100 bet would yield a $150 gain). You can probably guess which side offers better value.
http://www.economist.com/blogs/gametheory/2016/01/ranking-titans-basketball